New technological advances in the fields of alcohol safety and treatment sciences hold the promise of improving detection and control of alcohol impaired driving. Drivers with illegal blood alcohol concentrations (BAC) at or above the .08 g/dL level are involved in a fatal crash on average every 45 minutes every day. This study proposes to develop end-user risk analysis decision guides based on an exploration of two completely independent and objective technologies to scale the risk of impaired driving by DUI offenders: alcohol biomarkers from blood/urine, and the patterned record of serial BAC tests provided by drivers who use alcohol ignition interlock devices (IIDs). IIDs were introduced in 1986 to prevent an engine start if a driver had been drinking; they control the driver but also record and store BAC test data. Failed BAC tests strongly predict subsequent driving under the influence of alcohol (DUI) convictions. Blood and urine derived alcohol biomarkers also predict driver alcohol risk. Research has now documented convergence of these behavioral and biological methods of quantifying driver risk; the strongest biomarker predictors of driver risk are PEth, EtG, GGT, and %CDT. Alcohol biomarkers, which are measureable long after BAC has fallen to zero, allow for a more stable index of alcohol exposure. Alcohol biomarkers, IID BAC tests and psychometric assessments converge and can now be studied together. Today there are IID program laws in 48 states. Their use is growing rapidly (about 10% per year); by mid 2010, IIDs are used with an estimated 20% of all USA offenders convicted of DUI. It is an opportune time to study objective risk predictors more intensively. This proposal will examine independent source data on DUIs, and through 5 aims (yielding 5 papers) develop end user tools for detecting and codifying characteristics of the highest risk drivers who warrant more intensive control and monitoring. The basic research data to be explored has 534 IID drivers; 300 of them provided baseline blood for 6 major alcohol biomarkers (including GGT, PEth and %CDT). Subsamples of 90-150 DUI provided urine or hair for the other markers. Driver records, demographic and psychometric assessment data (DrinC, AUDIT, TLFB, TRI) are available for over 500. The 5 aims will find the best predictor combinations both at program entry and change over time, identify divergent subsets of drivers, evaluate sensitivity and specificity of predictors, then finally, synthesize, draft and solicit critiques for dissemination and communication of new insights via end user guides for authorities and providers who manage the DUI problem for society. Innovative methods will be joined from prevention, treatment and biomedicine to help inform about new approaches to help strengthen interlock programs. The proposed work will be supervised by senior investigators operating within the NIAAA funded Impaired Driving Center. The data set is completely unique. The product of this work has the potential to yield hard data for modeling driver risk and could serve as a basis for improving decisions about imposing more or less control on individuals.